search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.77k stars 7.96k forks source link

GIS and Darknet #4520

Open franjamonga opened 4 years ago

franjamonga commented 4 years ago

I would like to know if it is possible to extract the detected objects from the images, convert them to vector format. I would also like to know if the objects detected by being satellite images, if it is possible that they are still referenced with respect to the image, For example, if a car is detected, remove that car from the image and then use a GIS program so that when you open the image and the image of the car separately they are well positioned and referenced.

dselivanov commented 4 years ago

It is possible

franjamonga commented 4 years ago

@dselivanov Could you describe how the process would be or what files should I modify?

If I also wanted to train with tiff files and have the predicted image saved as tiff, what should I modify?

HagegeR commented 4 years ago

if you already know how tiff works, it's not that hard, build logic on top of the c++ or python binding 1) code that extract image from bounding box and image input. 2) code that updates the metadata for the image. if you want you can export output to a file to prototype faster using : -ext_output -dont_show -out result.json < data/train.txt at the end of your normal command

franjamonga commented 4 years ago

I would like to know what darknet files I must modify inside the src folder so that the predicted image is given in tiff format instead of jpg